I feel your pain. If there is something that I dislike more than a bad user interface, it is when it is changed to a different bad user interface.

Poor pharmaceutical companies, they have to jack up the price! Otherwise they wouldn't have such fabulous profit margins, or be able to spend such an incredible amount of money in marketing! Imagine the horror!

Or maybe people who are actually murdering innocents are terrorists?

As these videos led to a terrorist attack, I think charging those idiots with a felony is entirely appropriate.

That the felony in question is only related to the recording of the video is irrelevant to me. Al Capone was, after all, arrested for tax evasion.

The only thing we need for a simulation is the ability to approximate the solutions as well as we want. And the numerical methods do give us that.

The Wikipedia page about the Church-Turing-Deutsch thesis is garbage, I'm sorry for linking to that. I was in a hurry with a mobile phone and didn't read it, just went with blind trust in Wikipedia. A proper essay about the subject, written by a respected researcher, is here.

I can think of two objections to the idea that if post-humans come to exist they will be simulating us all the time:

1 - Post-humans will be themselves simulations (in the sense of running on a computer instead of brains, not in the sense of they themselves being simulated by post-post-humans). Therefore they will have strong ethical concerns about simulated people being created and destroyed at will.
2 - Since a huge chunk of their society will exist as a simulation, computing power will be valuable real state, unlikely to be wasted simulating boring people like us. If they would be interested in their history they would simulate the world once, store the interesting part of the results, and be done with it.

Go to higher order perturbation theory, the error goes to zero. Come on, I can't believe I'm being asked to provide evidence that one can solve differential equations numerically. You can, and that is why scientists look for differential equations to describe phenomena. This is a matter of scientific consensus.

The only objection against solving differential equations numerically is complexity, as one needs exponetial time to solve the ones for quantum systems. It is widely believed, however, that quantum computers will solve this problem. See the Church-Turing-Deutsch thesis.

The approximation I used was to consider the nucleus to be a point with infinite mass and charge +1, so the problem reduced to finding the ground state for two electrons in a 1/r potential, for which I used perturbation theory. I'm mystified about why do you think the simulation would fall apart as soon as I try to evolve it out of the ground state, or why quantum computers would fare any better at this: they would be solving the same equation.

Spoken like someone who has never actually tried to simulate those differential equations for something more complex than a hydrogen atom. I have actually solved Schrödinger's equation for a Helium atom numerically, it works fine. But you are right that one needs exponential time in general to simulate quantum mechanics in a classical computer.

This is not relevant, however, for either Hossenfelder's or Bostrom's arguments, as they don't hinge on the complexity of the simulation, but only on the possibility.

This is not her argument. Read her post again.

Your argument is even worse than hers. Just because we don't know what is the correct theory for quantum gravity it doesn't mean that we have no idea how complex it is going to be. For the universe to be non-simulatable we would need quantum gravity to be not merely very difficult to compute, but actually uncomputable. And there is zero evidence for that (no physical theory we have is uncomputable). In fact, the evidence we have points in the opposite direction, that the dimension of the Hilbert space of any bounded area is finite, because of things like the Bekenstein bound.

Yes, you can. Physicists solve Schrödinger's equation numerically in classical computers all the time. It is inefficient, of course, but efficiency plays no role in Hossenfelder's argument.

To simulate probabilistic outcomes you calculate the probabilities and use a random number generator.

Sabine Hossenfelder is conflating "living in a simulation" and "spacetime is discrete". For fucks sake, she is saying that we should see evidence of discretization via violations of Lorentz symmetry. Yes, this is true, a discrete universe is not compatible with the continuity of Lorentz transforms, but this has nothing to do with their simulatability. Lorentz transforms are trivial to simulate. Heck, all of physics we know can be simulated even in a classical computer, they are just differential equations.

Just because she's wrong it doesn't mean that the simulation argument is right, however. Personally, I think the simulation argument is uninteresting, because it is unfalsifiable, and therefore unscientific.

This might sound good, but is not true. I am a physicist, and I know lots of physicists. What happens is: as a student you make a lot of mistakes and get corrected all the time. As you learn stuff, you make less and less mistakes, and get corrected less and less. At the same time, you get used to knowing stuff that most people do not know. Even more, you get used to being right about stuff that most people are just mistaken about (like relativity, quantum mechanics, thermodynamics...). That has an effect on your psyche. As an active researcher, you still have the peer-review system to save your sanity: knowledgeable people criticizing your work keeps your ego in place. But after you retire, you just get prize after prize, and no scientific feedback anymore. Many physicists start believing that every brain fart they have is a genial idea, and that anyone that criticizes them doesn't know the first thing about what they are saying.

I don't think Hawking is going bananas. AI is a real risk that most people are blissfully unaware about. Hawking is just well-informed and thinking ahead.

This claim is just slander. If the claimed technology was in fact violating the second law of thermodynamics this wouldn't be a controversy, the paper would have been outright rejected and Goodenough would be hanging his head in shame.

Instead, what is apparently the problem is that there is some unaccounted-for surplus energy in the system, which must of course be explained if anyone is to take his experiment seriously.